LED Junction temperature: datasheet graph vs.temperature coefficient of voltage

[eee]

Hi guys!

I'm testing out some 12V Cree MK-Rs, and am having trouble calculating the junction temperature indirectly.

The datasheet shows the temperature coefficient of voltage as -8mV/°C

At room temperature (20°C), with a CC supply set to 1250mA, from cold, the LEDs' instantaneous Vf is 13.55V, settling to 13.05V after stabilising. According to the "temperature coefficient of voltage", this 0.5V reduction in Vf correlates to a 62.5°C increase in Tj (0.5V/0.008V). Since the room temperature was 20°C, Tj at the stabilised temperature would therefore be 82.5°C.


The problem is that the same datasheet states that at a Tj of 85°C and a drive current of 1250mA, the Vf should be 12.2V, as seen below:




Taking the 2.5C temperature difference to be negligible, there is still a 0.85V difference between the voltage drop on the graph, and the actual voltage drop.

Can anyone explain what's happening here? Is the LED running cooler than 85C because the Vf is far higher than the stated 12.2V on the graph, or is the calculated voltage drop from cold to saturated operating temperature a more accurate calculation of Tj?


Datasheet link if anyone needs it: http://www.cree.com/~/media/Files/Cree/LED%20Components%20and%20Modules/XLamp/Data%20and%20Binning/XLampMKR.pdf

[jpb]

The data sheet shows the coefficient as -8mV/C for the 12V variant as you say but the current is stated as 700 mA whilst you are measuring at 1250mA which is a lot larger.

It might be worth setting the current limit to 700mA (corresponding to around 11.7V according to the data sheet) and see if you get better agreement.

Doing any sort of detailed calculation based on "Typical" values from a data sheet is likely to be problematic, there might be a large device to device variation. As a simple check it is probably worth plotting the IV characteristics of your sample and seeing how that compares with the data sheet at least at a few points.

[dannyf]

Can anyone explain what's happening here?

You are comparing the behavior of a particular device vs. the typical. If you measure 100s/1000s of them, then there may be a good basis for comparison.

[eee]

The data sheet shows the coefficient as -8mV/C for the 12V variant as you say but the current is stated as 700 mA whilst you are measuring at 1250mA which is a lot larger.

It might be worth setting the current limit to 700mA (corresponding to around 11.7V according to the data sheet) and see if you get better agreement.

Doing any sort of detailed calculation based on "Typical" values from a data sheet is likely to be problematic, there might be a large device to device variation. As a simple check it is probably worth plotting the IV characteristics of your sample and seeing how that compares with the data sheet at least at a few points.

Can anyone explain what's happening here?

You are comparing the behavior of a particular device vs. the typical. If you measure 100s/1000s of them, then there may be a good basis for comparison.

Hi guys, thanks for the reply!

I have tested it at 700mA too, and the results were: cold start: 12.72V, after stabilisation: 12.46V. Still a 0.76V difference between expected (11.7) and measured. Theoretically, this 0.26V change = 32.5C increase above ambient, = a Tj of 62.5C

One thing to note is at 700mA and 1250mA drive currents, the forward voltages were 12.46V and 13.05V respectively, which is a range of 0.59V – the datasheet shows a range of 0.5V between those two drive currents, so at least this is kind of within expectations, although the theoretical Tj of both would be 62.5C and 82.5C, so not exactly constant.

The problem with your suggestion is that the datasheet graph shows figures for different drive currents and voltages ALL at 85C. Since I can't fix the junctions' temperature, my results won't compare to the graph...


I have 10 MK-Rs and checked if there's a difference in Vf at a tiny current (0.02A since they're not on a heatsink), and the Vf's were all 10.45V, one was 10.46V and one 10.47V.
When I first bought an MK-R 9 months ago (so a from a different batch), even with that I recall the Vf was at least more than 12.5V at 700mA, and for 1250mA it was in the 13V+ range at saturation, so it's not just this one LED that's much higher than expected...

P.S. Isn't the temperature coefficient of voltage supposed to be constant at all drive currents?

[jpb]

Maybe the actual diodes have a bit more contact/ohmic resistance than those measured for the data sheet - the data sheet values may have been for probed diodes before mounting.

The potential variation in forward current from the data sheet is pretty big e.g. the nominal 700mA at 85C could be as high as 1250mA.

Actually, given that 1250mA is the maximum forward current in the data sheet you perhaps affected the device a bit by driving that much through it for your first experiment.

Similarly the forward voltage at 700mA 85C could be as high as 14V according to the data sheet.

Another thing you could do is measure the temperature at the solder point if this is possible. The data sheet gives thermal resistance as 1.7 C/W typically so with 13V and 1.25A the difference in temperature between the junction and the solder point should be about 28C, so if the junction is around 85C the solder point should be around 57C. (You'd obviously have to let things reach equilibrium.)

[georges80]

LED Vf's will vary batch to batch, ESPECIALLY if they are from different bins.

This is one MAJOR reason that LED's need to be driven from a constant current supply. The Vf obviously also drops as the temperature increases. In addition, there is an 'aging' effect, typically the Vf will drop over the first few hundred hours of use.

So, you have Vf variations within a bin and more extreme between bins. You have Vf variations based on junction temps. You have Vf variations due to aging. Enough variables for you to play with?

cheers,
george.

[eee]

Thanks for your input guys. I thought I'd be able to use the Vf drop for some kind of indication at least, but I guess I'm just going to have to go on instinct and hope the junction temperature isn't too high! haha